Identification of Optimal Moderators for Time to Relapse

• Activity Number: 307 - Advanced Survival Analysis Tools for Statistical Learning from Complex Scientific Studies
• Type: Topic Contributed
• Date/Time: Wednesday, August 5, 2020 : 10:00 AM to 11:50 AM
• Sponsor: Lifetime Data Science Section
• Author(s): Yu Cheng* and Bang Wang
• Companies: University of Pittsburgh Department of Statistics and University of Pittsburgh
• Keywords: Counterfactual Outcomes; Matched Pair; Personalized Medicine; Smoking Cessation

Abstract

In personalized medicine, one is often interested in predicting treatment effect, which is the difference in counterfactual outcomes when the patient is assigned to an active treatment or control, based on patients’ individual characteristics. The evaluation of treatment effect becomes trickier when the outcome of interest such as time to relapse may be censored by the end of study. We adopt a pair design to identify personalized optimal treatment option in preventing relapse, and illustrate the methods using a randomized smoking cessation study.